Re: Localization issues: (WAS alpha v0.2)

["Martin J. Duerst" <duerst@w3.org>]

At 23:51 00/01/26 -0500, J. William Semich wrote:
> Thanks for bringing up this problem, Olafur...
> 
> At 09:45 PM 1/26/00 -0500, Olafur Gudmundsson wrote:
> 
> <big snip>
> 
> >We have choices such as
> >        limited but universal naming 
> >        rich but localized naming. 
> 
> We need to provide for both choices...
> 
> >Both can be supported, BUT we may end up with a system where two domain names
> >are needed, one from each choice. Thus if I want to use the domain 
> >"$B%b(JGu$B(Jcom" as my main domain, but that may not be universally expressable.  
> >I may need to include on my billboards and business cards the domain name
> >"ogud202.com" which is universal.  "ogud202.com" may only be a DNAME to 
> >"$B%b(JGu$B(Jcom" but that allows people that want to reach my site but do not
> >know how to or can not write Icelandic. 
> >
> 
> This is right on target as far as .NU Domain's implementation of Local
> Language domain names goes.
> 
> Can this fit into the "requirements" agenda for this WG? 

I think it's okay to require that it must be possible to have both
an internationalized and a 'downsized' (ASCII) name for a domain.
But I wouldn't agree to requiring that all domains have that.
I think this should be left to registrars, domain policies, and
so on.


> It's the number one issue for .NU Domain as we implement Local Language
> support worldwide. 
> 
> The .NU domain name registration system requires that any customer who
> wants to use our NUBIND service to register a Local Language domain name
> must also register an ASCII .NU domain name, which is conjoined to the
> Local Language name. An example would be domain.nu and dom$BgO(J.nu, although
> it is not required that the names be this closely related at all - only
> that one of them be ASCII. 

I think for the moment, this is a very reasonable policy.


> We will likely let customers add any number of different Local Language
> domain names to a single ASCII one, using any number of different character
> sets (so International companies can register several different versions of
> their domain name using various language's character sets). In effect, a
> single registration allows multiple character-set versions of a domain name
> to point to the same URL, email address, etc. 

Please don't use the term 'character set' any more on this list or its
documents. Unfortunately, it's ambiguous. Please either use 'character
encoding' ('charset') or something like 'collection of characters'.
I the above paragraph, I'm not clear which of them you are speaking
about, and it makes quite a difference.


> Sorry for going into implementation details, but it's important to consider
> how all this will actually work at the registration point and for end
> users. And solving the accessibility problem of URLs and email addresses
> that use Japanese Shift-JIS, for example, to users in the US using ASCII or
> Europe using ISO-8859-1, is one of those implementation problems that
> should be solved in the requirements, I'd say. The goal being to avoid
> segmenting the Internet.

I'm of course very clearly against segmentation. But there is one
big segmentation that already exists. It's the segmentation by
language. We won't be able to do miracles and assure that somebody
not speaking Islandic suddenly can read and understand Islandic
web pages. So why should we force somebody who has an Islandic-
only web site to have an ASCII-only domain name?

I think we should be able to assume that if somebody in the US
wants to look at Japanese web pages, they have a system/browser/
whatever that is able to handle Shift_JIS, or whatever.

There is the problem of bootstrapping, where as a DNS manager
you want to have a look at things before all of the system
is up, or without knowing the language and script of the web
site somebody tries to access. But for debugging,... an ASCII
equivalent won't be enough, and won't cover certain failures
unless it's always handed around together with the internationalized
one, which we can't ask for. So maybe we will need some conventions
to help some people for those debugging cases that can be
covered by such conventions, but I don't think we should
design our system to require a parallel ASCII registration
in all cases.


Regards,   Martin.



#-#-#  Martin J. Du"rst, World Wide Web Consortium
#-#-#  mailto:duerst@w3.org   http://www.w3.org